Real World Health Care Data Analysis , livre ebook

The correct bibliographic citation for this manual is as follows: Faries, Douglas, Xiang Zhang, Zbigniew Kadziola, Uwe Siebert, Felicitas Kuehne, Robert L. Obenchain, and Josep Maria Haro. 2020. Real World Health Care Data Analysis: Causal Methods and Implementation Using SAS ® . Cary, NC: SAS Institute Inc.
Real World Health Care Data Analysis: Causal Methods and Implementation Using SAS ®
Copyright © 2020, SAS Institute Inc., Cary, NC, USA
ISBN 978-1-64295-802-7 (Hard cover)
ISBN 978-1-64295-798-3 (Paperback)
ISBN 978-1-64295-799-0 (PDF)
ISBN 978-1-64295-800-3 (epub)
ISBN 978-1-64295-801-0 (kindle)
All Rights Reserved. Produced in the United States of America.
For a hard-copy book: No part of this publication may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, electronic, mechanical, photocopying, or otherwise, without the prior written permission of the publisher, SAS Institute Inc.
For a web download or e-book: Your use of this publication shall be governed by the terms established by the vendor at the time you acquire this publication.
The scanning, uploading, and distribution of this book via the Internet or any other means without the permission of the publisher is illegal and punishable by law. Please purchase only authorized electronic editions and do not participate in or encourage electronic piracy of copyrighted materials. Your support of others’ rights is appreciated.
U.S. Government License Rights; Restricted Rights: The Software and its documentation is commercial computer software developed at private expense and is provided with RESTRICTED RIGHTS to the United States Government. Use, duplication, or disclosure of the Software by the United States Government is subject to the license terms of this Agreement pursuant to, as applicable, FAR 12.212, DFAR 227.7202-1(a), DFAR 227.7202-3(a), and DFAR 227.7202-4, and, to the extent required under U.S. federal law, the minimum restricted rights as set out in FAR 52.227-19 (DEC 2007). If FAR 52.227-19 is applicable, this provision serves as notice under clause (c) thereof and no other notice is required to be affixed to the Software or documentation. The Government’s rights in Software and documentation shall be only those set forth in this Agreement.
SAS Institute Inc., SAS Campus Drive, Cary, NC 27513-2414
January 2020
SAS ® and all other SAS Institute Inc. product or service names are registered trademarks or trademarks of SAS Institute Inc. in the USA and other countries. ® indicates USA registration.
Other brand and product names are trademarks of their respective companies.
SAS software may be provided with certain third-party software, including but not limited to open-source software, which is licensed under its applicable third-party software license agreement. For license information about third-party software distributed with SAS software, refer to http://support.sas.com/thirdpartylicenses .


Contents

Contents
About the Book
What Does This Book Cover?
Is This Book for You?
What Should You Know about the Examples?
Software Used to Develop the Book’s Content
Example Code and Data
Acknowledgments
We Want to Hear from You
About the Authors
Chapter 1: Introduction to Observational and Real World Evidence Research
1.1 Why This Book?
1.2 Definition and Types of Real World Data (RWD)
1.3 Experimental Versus Observational Research
1.4 Types of Real World Studies
1.4.1 Cross-sectional Studies
1.4.2 Retrospective or Case-control Studies
1.4.3 Prospective or Cohort Studies
1.5 Questions Addressed by Real World Studies
1.6 The Issues: Bias and Confounding
1.6.1 Selection Bias
1.6.2 Information Bias
1.6.3 Confounding
1.7 Guidance for Real World Research
1.8 Best Practices for Real World Research
1.9 Contents of This Book
References
Chapter 2: Causal Inference and Comparative Effectiveness: A Foundation
2.1 Introduction
2.2 Causation
2.3 From R.A. Fisher to Modern Causal Inference Analyses
2.3.1 Fisher’s Randomized Experiment
2.3.2 Neyman’s Potential Outcome Notation
2.3.3 Rubin’s Causal Model
2.3.4 Pearl’s Causal Model
2.4 Estimands
2.5 Totality of Evidence: Replication, Exploratory, and Sensitivity Analyses
2.6 Summary
References
Chapter 3: Data Examples and Simulations
3.1 Introduction
3.2 The REFLECTIONS Study
3.3 The Lindner Study
3.4 Simulations
3.5 Analysis Data Set Examples
3.5.1 Simulated REFLECTIONS Data
3.5.2 Simulated PCI Data
3.6 Summary
References
Chapter 4: The Propensity Score
4.1 Introduction
4.2 Estimate Propensity Score
4.2.1 Selection of Covariates
4.2.2 Address Missing Covariates Values in Estimating Propensity Score
4.2.3 Selection of Propensity Score Estimation Model
4.2.4 The Criteria of “Good” Propensity Score Estimate
4.3 Example: Estimate Propensity Scores Using the Simulated REFLECTIONS Data
4.3.1 A Priori Logistic Model
4.3.2 Automatic Logistic Model Selection
4.3.3 Boosted CART Model
4.4 Summary
References
Chapter 5: Before You Analyze – Feasibility Assessment
5.1 Introduction
5.2 Best Practices for Assessing Feasibility: Common Support
5.2.1 Walker’s Preference Score and Clinical Equipoise
5.2.2 Standardized Differences in Means and Variance Ratios
5.2.3 Tipton’s Index
5.2.4 Proportion of Near Matches
5.2.4 Proportion of Near Matches
5.2.5 Trimming the Population
5.3 Best Practices for Assessing Feasibility: Assessing Balance
5.3.1 The Standardized Difference for Assessing Balance at the Individual Covariate Level
5.3.2 The Prognostic Score for Assessing Balance
5.4 Example: REFLECTIONS Data
5.4.1 Feasibility Assessment Using the Reflections Data
5.4.2 Balance Assessment Using the Reflections Data
5.5 Summary
References
Chapter 6: Matching Methods for Estimating Causal Treatment Effects
6.1 Introduction
6.2 Distance Metrics
6.2.1 Exact Distance Measure
6.2.2 Mahalanobis Distance Measure
6.2.3 Propensity Score Distance Measure
6.2.4 Linear Propensity Score Distance Measure
6.2.5 Some Considerations in Choosing Distance Measures
6.3 Matching Constraints
6.3.1 Calipers
6.3.2 Matching With and Without Replacement
6.3.3 Fixed Ratio Versus Variable Ratio Matching
6.4 Matching Algorithms
6.4.1 Nearest Neighbor Matching
6.4.2 Optimal Matching
6.4.3 Variable Ratio Matching
6.4.4 Full Matching
6.4.5 Discussion: Selecting the Matching Constraints and Algorithm
6.5 Example: Matching Methods Applied to the Simulated REFLECTIONS Data
6.5.1 Data Description
6.5.2 Computation of Different Matching Methods
6.5.3 1:1 Nearest Neighbor Matching
6.5.4 1:1 Optimal Matching with Additional Exact Matching
6.5.5 1:1 Mahalanobis Distance Matching with Caliper
6.5.6 Variable Ratio Matching
6.5.7 Full Matching
6.6 Discussion Topics: Analysis on Matched Samples, Variance Estimation of the Causal Treatment Effect, and Incomplete Matching
6.7 Summary
References
Chapter 7: Stratification for Estimating Causal Treatment Effects
7.1 Introduction
7.2 Propensity Score Stratification
7.2.1 Forming Propensity Score Strata
7.2.2 Estimation of Treatment Effects
7.3 Local Control
7.3.1 Choice of Clustering Method and Optimal Number of Clusters
7.3.2 Confirming that the Estimated Local Effect-Size Distribution Is Not Ignorable
7.4 Stratified Analysis of the PCI15K Data
7.4.1 Propensity Score Stratified Analysis
7.4.2 Local Control Analysis
7.5 Summary
References
Chapter 8: Inverse Weighting and Balancing Algorithms for Estimating Causal Treatment Effects
8.1 Introduction
8.2 Inverse Probability of Treatment Weighting
8.3 Overlap Weighting
8.4 Balancing Algorithms
8.5 Example of Weighting Analyses Using the REFLECTIONS Data
8.5.1 IPTW Analysis Using PROC CAUSALTRT
8.4.2 Overlap Weighted Analysis using PROC GENMOD
8.4.3 Entropy Balancing Analysis
8.5 Summary
References
Chapter 9: Putting It All Together: Model Averaging
9.1 Introduction
9.2 Model Averaging for Comparative Effectiveness
9.2.1 Selection of Individual Methods
9.2.2 Computing Model Averaging Weights
9.2.3 The Model Averaging Estimator and Inferences
9.3 Frequentist Model Averaging Example Using the Simulated REFLECTIONS Data
9.3.1 Setup: Selection of Analytical Methods
9.3.2 SAS Code
9.3.3 Analysis Results
9.4 Summary
References
Chapter 10: Generalized Propensity Score Analyses (> 2 Treatments)
10.1 Introduction
10.2 The Generalized Propensity Score
10.2.1 Definition, Notation, and Assumptions
10.2.2 Estimating the Generalized Propensity Score
10.3 Feasibility and Balance Assessment Using the Generalized Propensity Score
10.3.1 Extensions of Feasibility and Trimming
10.3.2 Balance Assessment
10.4 Estimating Treatment Effects Using the Generalized Propensity Score
10.4.1 GPS Matching
10.4.2 Inverse Probability Weighting
10.4.3 Vector Matching
10.5 SAS Progra